Porosity Detection Image Processing Algorithm for Aluminum Arc Welds

In aluminum GMAW processes, moisture in the atmosphere is dissociated into hydrogen by the arc plasma, and the resulting hydrogen readily dissolves into the weld pool due to the high hydrogen solubility of aluminum. Although most of the dissolved hydrogen escapes to the atmosphere during solidification, a portion that fails to diffuse out becomes trapped within the weld and forms porosity. These pores are critical defects that can significantly degrade the mechanical performance of welded joints; therefore, cross-sectional analysis and various non-destructive testing (NDT) techniques are commonly used for inspection. However, cross-sectional analysis provides only localized information and is insufficient for evaluating defect distribution along the entire weld. Moreover, X-ray radiography often contains noise and irrelevant artifacts, making accurate quantitative assessment challenging.</br>In this study, GMAW was performed on Al6061-T6 plates under conditions conducive to porosity formation, and X-ray radiographic images were acquired to capture defect information across the full weld region. MATLAB-based image processing techniques combined with a circular Hough transform algorithm were applied to quantitatively detect pore location, count, and diameter.